Research map / Learning theory and algorithms / Clustering
How to tell when a clustering is (approximately) correct using convex relaxations
Marina Meila
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| remove 60% of points closest to centers instead of outlier removal | SDP BTR bound · aspirin MD K=2 | BTR bound epsilon: .065 -> 0.047 | Hurt |
| remove 60% of points closest to centers instead of outlier removal | SDP BTR bound · aspirin MD K=2 | run time: 10h -> 42 minutes | Improved |
| increase dispersion sigma from 0.6 to 1.2 (ablation) | SDP BTR bound · K=4 unequal normal, n=200 | BTR bound epsilon: 0.00(0.00) -> 0.28 (0.08) | Hurt |
| increase n from 200 to 800 (ablation) | SDP BTR bound · K=4 unequal normal, sigma=1.2 | BTR bound epsilon: 0.28 (0.08) -> 0.21 (0.03) | No clear effect |
| use non-normal instead of normal clusters | SDP BTR bound · K=4 sigma=1.2, n=200 | BTR bound epsilon: 0.28 (0.08) -> 0.16 (0.06) | Improved |
| use SDP bound instead of spectral bound | spectral bound · K=4, n=1024 | bound tightness | Improved |
| place K=6 cluster centers along a line | spectral bound · K=6 clusters | bound quality | Hurt |
| decrease smallest cluster proportion p_min (ablation) | SDP BTR bound · unequal cluster sizes | BTR bound epsilon | Hurt |
| unequal vs equal cluster sizes | SDP BTR bound · K=4 | BTR bound epsilon | No clear effect |
Closest papers by meaning
- Convex Clustering: Model, Theoretical Guarantee and Efficient Algorithm (JMLR 2021 · similarity 0.87)
- Clustering with Semidefinite Programming and Fixed Point Iteration (JMLR 2022 · similarity 0.87)
- Efficient Constrained K-center Clustering with Background Knowledge (AAAI 2024 · similarity 0.85)
- Relax and Merge: A Simple Yet Effective Framework for Solving Fair $k$-Means and $k$-sparse Wasserstein Barycenter Problems (ICLR 2025 · similarity 0.85)
- Statistically Optimal $K$-means Clustering via Nonnegative Low-rank Semidefinite Programming (ICLR 2024 · similarity 0.85)
- Average Sensitivity of Hierarchical $k$-Median Clustering (ICML 2025 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
